JPS60233330A - Air-fuel ratio controlling apparatus for internal-combustion engine - Google Patents

Abstract

PURPOSE:To prevent fluctuation of the engine speed, by providing a means for learning a learning correction value from the correction value for closed-loop control of the air-fuel ratio at predetermined time intervals and a means for controlling the quantity of fuel supplied to an engine. CONSTITUTION:A air-fuel ratio detecting means (b) detects the air-fuel ratio of an internal-combustion engine (a) by detecting the concentration of a particular component contained in the exhaust gas of the engine. A means (c) calculates the correction value for closed-loop control of the air-fuel ratio from the conditions of the detected air-fuel ratio. A learning means (d) learns a learning correction value from the correction value for closed-loop control of the air-fuel ratio at predetermined time intervals. Further, a fuel supply means (e) controls the quantity of fuel supplied to the engine (a) according to the learning correction value and the correction value for closed-loop control. By employing such an arrangement, it is enabled to prevent fluctuation of the engine speed due to excessive frequency of learning control of the air-fuel ratio.

Description

TECHNICAL FIELD The present invention relates to an air-fuel ratio closed loop control device, and more particularly to an air-fuel ratio control device for an internal combustion engine having an air-fuel ratio learning control function.

BACKGROUND ART A technique is well known in which the air-fuel ratio is controlled in a closed loop to a target air-fuel ratio on the leaner side than the stoichiometric air-fuel ratio by detecting the concentration of a specific component in the exhaust gas using a concentration sensor called a lean sensor. In this type of linear air-fuel ratio control device, variations in the characteristics of the lean sensor itself and non-uniformity of the air-fuel mixture supplied to each cylinder are relatively large, so closed-loop control is performed to keep the deviation from the target air-fuel ratio as small as possible. The calculation period for the correction amount is shortened, and the width of each change in the correction amount is made small. For this reason, the closed-loop control correction amount is frequently reversed in its increasing/decreasing direction.

Learning of the air-fuel ratio is normally performed every time the closed-loop control correction amount reverses its increase or decrease several times, for example, every five times. For this reason, in the lean air-fuel ratio control device as described above, learning is performed too frequently, and at idle, the learning value fluctuates due to the movement K of the closed-loop control correction amount, and this fluctuation in the learning value causes hunting in the rotational speed. There is a problem with this.

Further, when the operating state of the engine is divided into a plurality of divisions and learning values are obtained for each division, the following inconveniences may occur if the learning is performed in accordance with the movement of the waveform of the closed-loop control correction amount.

It is difficult to perform learning unless the closed-loop control correction amount increases or decreases and is reversed.On the other hand, since the operating state of the engine changes rapidly, the values learned in one operating state category may be different from those in other categories. There is a risk that it will be recognized as something. If the segmented area is learned incorrectly in this way,
This results in deterioration of driving characteristics and exhaust gas purification characteristics.

OBJECTS OF THE INVENTION Accordingly, the present invention solves the above-mentioned problems of the prior art, and its purpose is to prevent the rotational speed from fluctuating due to too frequent learning. Another point is to prevent the occurrence of a situation in which correct learning cannot be performed due to too long learning intervals.

Structure of the Invention The structure of the present invention that achieves the above-mentioned object will be explained with reference to FIG. , means C for calculating a closed-loop control correction amount for the air-fuel ratio according to the detected air-fuel ratio state, means d for learning a learning correction amount from the closed-loop control correction amount at predetermined time intervals, and the closed-loop control correction amount and learning. The present invention is characterized by comprising means e for controlling the amount of fuel supplied to the engine a in accordance with the correction amount.

Example The present invention will be described in detail below using Examples.

FIG. 2 schematically shows an internal combustion engine whose fuel injection is controlled by a microcomputer as an embodiment of the present invention. In the figure, 10 is an intake pipe connected to an air cleaner 12, and 14 is a throttle valve provided in the middle of the intake pipe 10.

The throttle valve 14 controls the intake air flow rate in conjunction with an accelerator (not shown).

The throttle switch 16 is connected to the rotating shaft of the throttle valve 14, and closes when the throttle valve 14 is in a fully closed state (idle position) to generate a signal to that effect. This throttle fully closed signal is sent to the electronic control unit) (fi
1: Send it to the input/output (Ilo) port 18b of the CU) 18 and use it to the appropriate size.

A surge tank 20 connected to the intake pipe 10 is provided with a pressure sensor 22 that detects the absolute pressure inside the intake pipe. The pressure sensor 22 outputs a voltage corresponding to the detected intake pipe internal pressure, and this output voltage is [DCU18
The signal is sent to an analog-to-digital (A/D) converter 18a.

The surge tank 20 is connected to an intake manifold 24, and the intake manifold 24 is connected to the combustion chamber 26 of each cylinder.
connected to. Fuel injection valve 2 is installed in the intake port of each cylinder.
8 are installed respectively. ECU: 1.8 Injection signals are sent to each fuel injection valve 28 via the I10 port 18b and the drive circuit 18c, so that each fuel injection valve 28 is intermittently opened and closed, and the fuel supply system (not shown) is activated. Intermittently injects pressurized fuel sent from

A lean sensor 32 is attached to the exhaust pipe (or exhaust manifold) 30 and generates a current as shown in FIG. 3 in accordance with the oxygen Jiv4 concentration in the exhaust gas. The structure, characteristics, usage examples, etc. of such lean sensor 32 are as follows.
It is widely known from Japanese Patent Application Laid-Open No. 58-143108. The output of the lean processor 32 is subjected to current-to-voltage conversion by a conversion circuit 18d in the ECU 18, and then applied to a voltage converter 18a.

Crank angle sensors 36 and 38 are attached to the distributor 34. From these crank angle sensors 36 and 38, the engine's illustrated crankshaft is 30
A pulse signal is output every time the motor rotates by 720° and 720°, and is applied to the output port 18b of the FJCU 18.

The ECU 18 includes a central processing unit (CPU) 18e, a random access memory (
RAM) 18 f, and read-only memory IJ
(ROM) 18g, etc. A/1)
The converter 181L also has a multiplexer function, and selects the voltage corresponding to the output voltage of the pressure sensor 22 or the output current of the lean sensor 32 according to the instruction signal given from the CPo 18e at predetermined time intervals, Convert to 2-shot signal. The obtained two signals, i.e., -Y and Lean processor 320 output LNS representing the intake pipe internal pressure PM.
The r-event corresponding to H is stored in the RAM 18f.

/P pulse signals from crank angle sensors 36 and 38 are I
/(It is sent to the CPU J 80 via the port 18b and is used for cylinder discrimination, crank angle position classification, rotation speed calculation, etc. For example, by measuring the time required for the crank lly to rotate 180 degrees The rotational speed NE can be known.The NE obtained in this way is stored in RAM18.
It is stored in f.

The ROM 18g stores in advance a control program, a function table, etc., which will be described later.

Next, the operation of this embodiment will be explained using a flowchart.

FIG. 4 shows a control program for calculating the fuel injection pulse width TAU, and the CPU I, p, and CPU 8e execute this processing routine at every predetermined crank angle, for example, every 180° crank angle, during the main routine.

In Stella f100, basic /4 is calculated from the data of rotation speed Ng and intake pipe pressure PM stored in RAM18f.
'Russ width TP is observed. To calculate this basic pulse width TP, use the NE pre-stored in the ROM 18g.
, PM and TP function tables are used. In the next Stella 7″lO1, the fuel injection/selse width TAU is this basic/elusse width TP, the air-fuel ratio feedback correction coefficient FA
F, IJ-ton correction coefficient KLEAN, learning correction amount KGi in its operating state classification l, and other correction coefficients α
, β can be calculated from the following equation.

TAU=TP −FAF −KLEAN −KGi・α
1βFAF is a coefficient for performing closed loop control of the air-fuel ratio, and is calculated by the processing routine shown in FIG. In the case of closed loop control, FAF=1.0 is fixed.

KLEAN is a correction coefficient for setting the target air-fuel ratio to a value closer to IJ than the stoichiometric air-fuel ratio, and is variably controlled according to the operating state. If the target air-fuel ratio is the stoichiometric air-fuel ratio,
KIJAN=1.0 is set. KGl is a correction amount for performing learning correction of the air-fuel ratio, and the operating state of the engine is divided into a plurality of regions, and learning is performed for each classification region. In areas where learning control is not performed, KGi is set to 1.0. Note that the learning of KGi with and is performed in the processing routine shown in FIG.

In the next Stella 7'102, the requested fuel injection pulse width TAU is stored in the RAM 18f.

During the interrupt processing routine executed at each predetermined crank angle position of each cylinder, the injection start time and injection end time are determined from this fuel injection/glucose width TAU, and during these times the injection signal is transmitted to the corresponding port of the I10 port 18b. Furthermore, it is output to the cylinder position. As a result, fuel injection is performed as described above.
- FIG. 5 is a processing routine regarding the output LNSR of the lean sensor 32. The CPU'18e executes this processing routine in the middle of the main routine at predetermined intervals.

In step 200, the output LNSR of the lean sensor 32
and a comparison reference value IR to determine whether the current air-fuel ratio is on the rich side or on the one side of the target air-fuel ratio determined by the comparison reference value IR.

This comparison reference value IR may be a constant value, but by making it a variable value according to the correction coefficient KLEAN, the target air-fuel ratio can be variably controlled according to the engine operating state.

If LNSR>IR, that is, if it is on the lean side, proceed to Stella 7'201 and reset the flag XAF to @0''.This flag This flag is used in the FAF calculation routine shown in Figure 6.LNS
If R≦IRQ, that is, if it is on the rich side, Ste.
Go to step 202 and set XAFt-@1''.

FIG. 6 is an example of a processing routine for calculating the air-fuel ratio feedback correction coefficient FAF. There are various other methods for calculating F'AF. The processing routine shown in FIG. 6 may be executed following the processing routine shown in FIG. 5, or may be executed separately at predetermined intervals as a main routine or the like.

First, in step 300, it is determined whether closed-loop control execution conditions are satisfied. While starting the engine, increasing the warm-up amount,
・Since the closed loop condition does not hold when the amount of water is increased, etc.,
Proceed to step 301 and set FAF=1.0. Next, in Stella f302, the initial value "255-160" is set to the count value CTR for determining the timing to perform the learning process.
”(=95) is given.

If the closed loop condition is satisfied, the process proceeds to step 303, where it is determined whether the flag XAF is XAF=1. If XAF'=1, it is hit, and if it is on the rich side, steps 304 to 308 are performed. In step f304, the gap 7° flag CAFL used in steps f309 to f313 is reset to "0". Next Stella 7'
In step 305, it is determined whether the skilag flag CAFR is 0#. When shifting from the IJ-on side to the rich side for the first time, CAFR=0, so the process goes to step 306, performs the skip processing routine shown in FIG.
Stella f307 after greatly decreasing (SKPI only)
The flag CAFR is set to ``1'' at step 305. When Stella f305 is executed next, the flow advances to step 308. In Stella f308, FAF is decremented by 1.

Note that SKP is a one-digit constant, and 5KPI is Kl
Yorikana 9 is selected as a large value. 5KPl is for performing processing to greatly reduce FAF, that is, skip processing, when it is determined that the air-fuel ratio has shifted from lean to rich with respect to the target air-fuel ratio. Also, K1 is F
This is for integral processing to gradually reduce AF.

When XAF=O is 1, that is, on the lean side, the processing of Stella f309 to f313 is performed. These processes are
The process is similar to steps 304 to 308 described above except that FAF is increased by SKP or :. Incidentally, in step 7'311, the skip processing routine shown in FIG. 8 is performed. The FAF thus obtained is stored in the RAM 18f at step f314.

FIG. 7 shows the skip processing routine of step 306 in FIG. 6 in detail.

In Stella, 7' 306 a, F'AF' immediately before the skip used in the skip processing routine on the 311 side.
Store it as 2iFAF L and peel it. next step 30
6b is the arithmetic mean value of the FAF before this Sugi 7Djθ and the FAFR representing the FAF just before the previous Skip 2° (this is Stella f311 a Temera Fang in Figure 8)!
The calculation of 'AP' AV is performed.The operation of hitting is performed.In the next Stella f:306c, FAF is SKP,
The skip processing is actually performed by decreasing the amount by t.

In the next step 306d, the count value CTR is incremented by one, and when this value CTR is determined to be CTR=256 and the next Stella 7' 306e, Stella f
306 and 306g are processed. That is, count 1
When the direct CTR is CTR=255 and this ski, f processing routine is executed, steps 306f and 3
Process 7 of 06g will be performed. Step 306
In f, the count value CTR is set to the initial value "255-32# (=
223) is given, and the flag FX1 is set to 11' in Stella 7'306g. This flag li'
I is a flag requesting execution of learning.

Figure 8 shows the skip processing routine of Stella'/'311 in Figure 6 in detail.
It is exactly the same as that shown in FIG. 7 except that FL and FL are exchanged.

FIG. 9 is a processing routine that is executed every predetermined time (32 mlec) and detects when CTR is incremented to count (I and CTR=256).

First, in Stellar f400, CTR is incremented by one. In the next step 401, it is determined whether 1cTR=256 or not, and only when CTR=256, the process advances to step 402 and this CTR is set to 255'''.

In this way, according to the processing routine of FIG. 9, 32m se
CTR is incremented by one every c.

Since the initial value "'255-32" is given to CTR in steps 306f and 311f of FIGS. 7 and 8,
Every time the processing routine in FIG. 9 is executed 33 times, CTR=
255 As a result, Stella f306g and 311g of the processing routines in Figures 7 and 8 are executed and the FX is executed.
,=1. That is, 32X32msec (=10
24m5ec) Every o, < FX, will be set to 11" at the time of key f processing. As will be described later, l'
If 4=l, learning of KGi is performed, and when learning is performed, FX is reset to "+".

FIG. 10 shows a processing routine that is executed during the main routine.
When it is determined that the state has changed from the open state to the fully closed state, the process proceeds to step 501, and the count value CTR is set to the initial value "'255-".
160'' (=95) is given.

In this way, when the throttle valve 14 changes from open to fully closed during deceleration, give a large initial value to CTR and wait for a while (
32X160msec=5120msec, about 5 seconds),
Prevents learning from occurring. This is because the air-fuel ratio changes from lean to rich for a while after the throttle valve is closed. Also, the reason why a similar large initial value is given to CTR in step 302 of FIG. 6 is because the characteristics of the lean sensor 32 are not stable for a while after changing from open loop to closed loop control.
This is to prevent learning from occurring during that time.

FIG. 11 shows the air-fuel ratio learning correction at-KG for each operating range.

This is a processing routine for setting ~KGn. The operating range is divided into a plurality of regions depending on the intake pipe internal pressure PM and whether or not the throttle is fully closed, and a learning correction amount is determined for each operating range. First, in step f600, it is determined whether roof control is currently being performed. The process proceeds to step 601 only when the control is not closed-loop control, that is, when the control is closed-loop control. In step 601, it is determined whether the flag FX is @On or not, and if FX1=00, the following learning process is not performed. If FX, = 1, approximately 1 second has passed since the last learning, or when the throttle valve is fully closed or closed! The following learning process is performed assuming that approximately 5 seconds have passed since the start.

In Step 02, the FAF determined by the processing routine in Figure 6 (steps 306b and 311b in Figures 7 and 8)
It is determined whether AV is 0.9 or more, and in the next Stella f603, it is determined whether FAFAV is 1.1 or less. FAFAV (step 60 if 0,9
Proceed to step 4 and set the learning correction amount KGi for this driving range to a constant value γ.
Only a decrease can be avoided. That is, processing of KGi4-KGi-γ is performed. Also, 1.1 (for FAF'AV, Stella f6
Proceed to step 05 and increase KGi by γ.

That is, processing of KGi4-KGi 10γ is performed. Furthermore, o
If 9≦FAFAV≦1.1, proceed directly to Stella f607. Stella f604h or Stella 7°605 KG
When the arithmetic processing of i has been performed, FX is reset to 60# in Stella 7'606. In step 607, the learning correction amount KGi obtained as described above is written into each storage location of the RAM 18f.

In the lean air-fuel ratio control device, F'AF is skipped very quickly as shown by arrow A in FIG. 12 (4).

For this reason, if learning is performed every five skids as in the past, the learning correction amount KGi will fluctuate greatly (first
Figure 2 (g))). As a result, as shown in the 12th section, the rotational speed during idling causes hunting and large fluctuations, resulting in the inconvenience of I-run. In addition, the first
In Figure 2, B indicates the learning dead zone.

On the other hand, in this embodiment, learning of the learning correction amount KGi is performed at predetermined time intervals (approximately 1 second) as shown in FIG. Even if this happens, learning will not be performed in response to this, and therefore, there will be no inconveniences such as fluctuations in the learning value and fluctuations in rotational speed during idling. Also, conversely, F
Even if AF skipping does not occur, learning is performed at regular time intervals, so a correct learning correction amount can be obtained in each driving state classification, and deterioration of driving characteristics and exhaust gas purification characteristics can be prevented.

Note that the method for performing the learning process at constant time intervals is not limited to the above-mentioned Glodarum example and embodiment, and it is clear that various other methods can be used.

Furthermore, the embodiment described above is an intermediate step to a lean air-fuel ratio closed-loop control device that performs closed-loop control of the lean air-fuel ratio to a value on the IJ-n side of the stoichiometric air-fuel ratio.
The present invention can obtain similar effects even when applied to a normal air-fuel ratio closed loop control device that controls the air-fuel ratio to the stoichiometric air-fuel ratio.

Effects of the Invention As described in detail above, according to the present invention, since the learning correction amount is learned at predetermined intervals, it is possible to prevent the rotational speed from fluctuating due to too frequent learning. There are no inconveniences such as inability to learn properly due to too long study intervals.

[Brief explanation of drawings]

Figure 1 is a configuration diagram of the present invention, Figure 2 is a schematic diagram of an embodiment of the present invention, Figure 3 is a characteristic diagram of a lean sensor, and Figures 4 to 1.
Figure 1 shows the flow of part of the control program = +-, y -h
, FIG. 12 is a diagram for explaining the problem of the prior art, and FIG. 13 is a diagram for explaining the operation of the above embodiment. 10... Intake pipe, 12... Air cleaner, 14...
...Throttle valve, 16...Throttle switch, 1
8...ECU, 18a...A//LD converter, I
8 b・-1/? ) port, 18c... drive circuit,
18d...conversion circuit, 18e...CPU, isr.
...RAM, 18 g...ROM, 22...Pressure sensor, 24...Intake manifold 526...Combustion chamber,
28...Air injection valve, 30...Exhaust pipe or exhaust manifold, 32...Lean sensor, 34...Distributor, 36.38...Crank angle sensor. 1 patent agent Toyota Motor Corporation Patent application agent Akira Aoki Patent attorney Kazuyuki Nishidate Patent attorney Kenzo Hiraiwa Patent attorney Akira Yamaguchi Patent attorney Masaya Nishiyama Figure 3 Figure 4 Figure 5 Go to step 307 8 Figure Step 312 from step 310\

Claims (1)

[Scope of Claims] 1. Means for detecting the air-fuel ratio state of the engine by detecting the concentration of a specific component in exhaust gas, and means for calculating a closed-loop control correction amount for the air-fuel ratio according to the detected air-fuel ratio state. , comprising means for learning a learning correction amount from the closed-loop control correction amount at predetermined time intervals, and means for controlling the amount of fuel supplied to the engine according to the closed-loop control correction amount and the learning correction amount. Air-fuel ratio control device for internal combustion engines. 2. The air-fuel ratio control device according to claim 1, wherein the learning means stops learning the learning correction amount for a predetermined time from the time when the throttle valve is fully closed.